Table 7: Model 7: Likelihood of Double Contact With 3 Dichotomous Empowerment Variables
Variable Coefficient
Standard
Error
Partisan Strength
0.571
(0.317*)
Age -0.008
(0.005)
Homeowner 0.014 (0.159)
Education -0.062 (0.101)
Income 0.001
(0.002)
Gender -0.152
(0.135)
Married 0.022
(0.333)
Native 0.172
(0.149)
Language -0.104
(0.094)
Florida 0.336
(0.226)
California 0.061 (0.241)
Illinois 0.409
(0.215*)
New York
-0.079
(0.261)
1 Latino (0,1)
0.265
(0.199)
2 Latinos (0,1)
0.396
(0.232*)
3 Latinos (0,1)
0.327
(0.211)
Group Size
-0.001
(0.004)
Constant -20.841
(0.758**)
†Dependent Variable is Exclusively Partisan AND Non-Partisan Contact
††*-Significant at 90%, two-tailed; **- Significant at 95%, two-tailed
In order to provide an overall summary of the effects of particularly relevant variables
across all our models, we calculated the change in the probabilities of being contacted as we
changed the values of certain variables. Specifically, we are interested in what the probability of
contact for the “most empowered” Latinos relative to the “least empowered”. Across many of
the models, we saw strength of partisanship and homeowner as significant predictors of some
form of contact. To calculate the predicted probabilities of contact for each model we set these
two variables to their minimum and maximum values to predict low and high probabilities of
contact, respectively. In addition, we are also interested in the effects of empowerment and
group size upon the probability of contact. To calculate predicted probabilities for each model,
we also set group size to its minimum and maximum values. In models 1-3, we set the
empowerment variables to their minimum and maximum values. In models 4-5, we only set the
indicator variable for One Latino Representative to its minimum maximum values and hold all
other measures of empowerment at its mean values. In models 6-7, we set the indicator variable
for Two Latino Representatives to its minimum and maximum values and hold all other
empowerment measures to their mean values. The predicted probabilities with these
calculations are presented in the table below.
Barreto, Espino, Pantoja and RamĂrez / APSA 2003 / 18
Convention